Elastomeric substrates with embedded stiff platforms for stretchable electronics

CitationRomeo, Alessia, Qihan Liu, Zhigang Suo, and Stéphanie P. Lacour. 2013. “Elastomeric Substrates with Embedded Stiff Platforms for Stretchable Electronics.” Applied Physics Letters 102 (13): 131904. https://doi.org/10.1063/1.4799653.

Permanent linkhttp://nrs.harvard.edu/urn-3:HUL.InstRepos:41461298

Terms of UseThis article was downloaded from Harvard University’s DASH repository, and is made available under the terms and conditions applicable to Other Posted Material, as set forth at http://nrs.harvard.edu/urn-3:HUL.InstRepos:dash.current.terms-of-use#LAA

Share Your StoryThe Harvard community has made this article openly available.Please share how this access benefits you. Submit a story .

Accessibility

Elastomeric substrates with embedded stiff platforms for stretchableelectronics

Alessia Romeo,1,a) Qihan Liu,2,a) Zhigang Suo,2 and St!ephanie P. Lacour1,b)1Centre for Neuroprosthetics, EPFL|STI|IMT/IBI|LSBI, 1015 Lausanne, Switzerland2School of Engineering and Applied Sciences and Kavli Institute for Bionano Science and Technology,Harvard University, Cambridge, Massachusetts 02138, USA

(Received 5 February 2013; accepted 21 March 2013; published online 2 April 2013)

Stretchable electronics typically integrate hard, functional materials on soft substrates. Here wereport on engineered elastomeric substrates designed to host stretchable circuitry. Regions of a stiffmaterial, patterned using photolithography, are embedded within a soft elastomer leaving a smoothsurface. We present the associated design rules to produce stretchable circuits based on experimentalas well as modeling data. We demonstrate our approach with thin-film electronic materials. The“customized” elastomeric substrates may also be used as a generic elastic substrate for stretchablecircuits prepared with alternative technologies, such as transfer-printing of inorganic, thinneddevices.VC 2013 American Institute of Physics. [http://dx.doi.org/10.1063/1.4799653]

Stretchable electronics, i.e., integrated circuitry that canreversibly expand and relax, are hybrid systems, combiningmechanically disparate, soft and hard, materials within a sin-gle structure.1 In most cases, the carrier substrate is an elasticor viscoelastic polymer, e.g., silicones or polyurethanes, char-acterized by low elastic modulus (E< 10MPa), large ductility(elongation at break>100%), Poisson’s ratio ! close to 0.5,and thickness in the 10lm to 1mm range. By contrast, elec-tronic device materials, used either in thin-film or thinnedforms, are stiff (elastic modulus in the GPa range), brittle(fracture strain<5%), and thin (thickness<1lm). The mostcommon design of stretchable circuitry is to produce apixelated2–5 or meshed6–8 macroscopic structure. The “pixels”or nodes are made of hard materials. A soft elastomeric sub-strate supports the mechanically rigid nodes and isolates themfrom the applied macroscopic strain. Figure 1(a) shows across-sectional view of the structure: non-deformable plat-forms hosting fragile electronic materials are distributed ontop of the rubbery substrate and are interconnected with elas-tic wiring.2,9 An elastic encapsulation (not shown in the draw-ing) can also be added.10

Several advanced stretchable circuits, prepared using thisdesign, have recently been demonstrated and applied to large-area electronics,6 biomedical wearable interfaces10 andimplantable circuitry.11 These circuits are fabricated withcomplex, multi-step, multi-carrier processing. Active devicematerials are first deposited and patterned on a rigid or plasticsubstrate, which in turn is machined into a thin mesh definingthe rigid nodes, and subsequently transferred onto the elasticmatrix. Complex wiring technology, based on thick compositeelastomers,7 2D,8 or 3D10 meandering structures, is requiredto interconnect electromechanically the stiff nodes. Despitethe latest demonstrations of stretchable circuits, this hard-on-soft, pixelated design suffers from large strain concentrationat the rigid-to-elastic transition zones, which often limits thelong-term performance of the stretchable circuit.8 Here we

introduce an alternative approach where the pixelated circuitsare manufactured directly onto a planar but mechanicallyengineered heterogeneous elastic substrate. We further pres-ent the associated design rules to produce stretchable circuitry.To ensure minimal or no strain in a layer or multilayer of brit-tle electronic materials (which should not be strained beyonda critical strain of !0.5% to remain crack-free upon stretch-ing), we pattern them above built-in, rigid platforms prelimi-nary distributed and embedded within the soft substrate.Figure 1(b) presents a cross-section of the optimized design.Throughout the study, we assume that the device island is sig-nificantly thinner than the engineered substrate. To maximizethe available surface area of the active circuits and to mini-mize the strain concentration at the rigid-to-elastic transitionzones, we optimize, using micromechanics modeling, thedesign parameters of the substrate, namely the embedded rigidplatforms’ diameter D, thickness h, and spacing S, the thick-ness of the substrate t, and the diameter of the device islandsd. Furthermore, we demonstrate the engineered elastomericsubstrate is compatible with standard, additive thin-filmprocessing, thereby, promises low cost manufacturing andscalability.

The engineered substrate is manufactured using polymerspin-coating and photolithography. First, stiff platforms areprepared using a photopatternable polymer with a highYoung’s modulus (E> 1GPa), e.g., SU8 epoxy resist. Theirdimensions and density are programmed with the lithographymask and the polymer spin speed. The platforms are 1mm di-ameter, 10–90lm thick, and organized in a square pattern of2–10mm inter-platform distance. After curing, the platformsare fully embedded in 100lm thick silicone rubber (Sylgard184, Dow Corning) cured at 80 "C for 24 h (Figure 1(b)). Theengineered substrates are then released from the Si wafers af-ter anodic dissolution of a metallic releasing layer.

When the platform is significantly stiffer than the sur-rounding silicone matrix, the elastomer volume immediatelyabove the platform is little strained when the matrix is macro-scopically stretched; therefore, brittle materials deposited onthe corresponding top surface are not extensively stretched.Figure 2 provides an experimental illustration of our

a)A. Romeo and Q. Liu contributed equally to this work.b)Author to whom correspondence should be addressed. Electronic email:stephanie.lacour@epfl.ch

0003-6951/2013/102(13)/131904/5/$30.00 VC 2013 American Institute of Physics102, 131904-1

APPLIED PHYSICS LETTERS 102, 131904 (2013)

approach. We deposited aluminum oxide (Al2O3) thin-filmdisks onto plain (Figure 2(a)) and engineered (Figure 2(b)) sil-icone substrates. The Al2O3 disks (150 nm thick) are intercon-nected with stretchable thin-film gold conductors (chromium/gold (3/30 nm)) patterned directly onto the elastomeric sub-strate.12,13 The engineered substrate is prepared with SU8photoresist platforms (Gersteltec GM 1070, ESU8¼ 4GPa,D¼ 1mm, h¼ 50lm) embedded in polydimethylsiloxane

(PDMS) rubber (EPDMS¼ 1MPa, t¼ 100lm). Before stretch-ing, the thin-films on top of the rubbery substrates are crack-free (spontaneous wrinkles are often observed in the goldfilm). Upon stretching to 20% strain, Al2O3 disks on bulkPDMS crack at low strain (e< 5%). On the other hand, Al2O3

disks patterned on PDMS above the SU8 platforms do notcrack and sustain repeated mechanical loading reliably.Figure 2(c) displays the electrical resistance of a gold conduc-tor running across 4 un-cracked Al2O3 islands during 100cycles to 20% applied strain. The outer envelop of the resist-ance versus cycling is similar to those reported in Refs. 13and 14.

We monitored the surface strain across the engineeredsubstrate during tensile loading to 20% strain in order to ana-lyze the strain distribution at the rigid-to-elastic interface.No delamination of the SU8 platforms from the surroundingPDMS is observed during cycling to 20% strain; the top sur-face of the PDMS is smooth.

According to fracture mechanics, the critical strain for abrittle film to fracture scales with the film thickness d asefracture / 1=!d.15 For example, a 1 lm thick SiOx film frac-tures at about 1% strain, whilst a 7 nm thick film can sustainstrains up to 7%.16 On the other hand, we define the value of0.5% as the critical strain not to be exceeded in the materialabove the platforms. This more realistic criterion is intendedto ensure that the most fragile materials can be protectedfrom cracking when integrated on a stretchable substrate.We define a “safe” top surface area, AS as the isolated sur-face where the resulting strain er remains lower than the

FIG. 2. Stretchable alumina disks onengineered elastomeric substrate. Top-view optical images recorded during astretch cycle to 20% applied strain of (a)a 150 nm thick, 1mm diameter, Al2O3

disk deposited onto 0.1mm thick, bulkPDMS substrate and interconnected witha thin metal conductor (5/25/5 nm Ti/Au/Ti thin films); scale bar: 200lm; (b) a150 nm thick, 0.75mm diameter, Al2O3

disk deposited above 1mm diameter,50lm thick SU8 platform embedded inthe 0.1mm thick PDMS substrate andinterconnected with a thin metal conduc-tor (5/25/5 nm Ti/Au/Ti thin films); scalebar: 200lm. (c) Electrical resistance as afunction of time of the metallic conduc-tor running across 5 Al2O3 disks on engi-neered substrate during 100 stretchcycles to 20% strain.

FIG. 1. Architecture of stretchable electronic circuits. (a) Cross-sectional viewof a previous design. Stiff (nondeformable) platforms are deposited on top ofthe stretchable substrate. Electronic devices are built or transferred onto theplatforms and interconnected with elastic wiring. (b) Cross-sectional view ofthe alterative design. Stiff platforms are embedded within the stretchable sub-strate. Electronic devices are manufactured directly onto the flat, elastomersurface, above the stiff platforms, and interconnected with elastic wiring.

131904-2 Romeo et al. Appl. Phys. Lett. 102, 131904 (2013)

critical strain 0.5%, independent of the applied macroscopicstrain eappl.

Increasing the SU8 platform thickness h maximizes As.A 90 lm thick platform embedded in 100 lm thick PDMSsubstrate ensures nearly 95% of the surface above the SU8platform is strain-free. However, sharp strain peak at theedge of the platform and out of plane distortion of the sur-rounding PDMS are observed with the very thick platforms.Figure 3 shows surface strain maps and strain profiles meas-ured across the engineered substrates as a function of thespacing between 50 lm thick and 1mm diameter SU8 plat-forms. The strain across the bare surface of the elastic sub-strate was mapped under uniaxial stretching using anextensometer equipped with a Digital Image Correlationsystem (LIMESS Messtechnik und Software GmbH, Istra4D v4.3.0 software). The strain immediately above theSU8 remains low, close to 0% independently on the inter-platform distance. However, the strain in between the plat-forms and the peak strain at the edge of the SU8 platformsincrease with decreasing spacing. 3D computed finite-element profiles of the same architectures, shown in Figure3(c), match well with the experimental data.

Experimental strain mapping provides valuable informa-tion on the lateral design of the engineered substrate, i.e., the

platforms diameter and spacing, and the corresponding strainin-between the platforms. However, the experimental set-upfails to capture accurately the fine details of the rigid-to-elas-tic region above the edge of the platforms while the sampleis stretched. We, therefore, focused on using microme-chanics modeling to further optimize the design on the engi-neered substrate. We aim at predicting where the higheststrain appears, defining how far inside the platform it is safeto pattern the devices, i.e., define the “safe” surface area As

(relative to the platform surface area, pD2

4 ), and clarifyinghow these factors are influenced by the structure geometry.The strain along the top surface of the substrate takes the fol-lowing form:

ex

t

! "¼ f

x

t;g

t;D

t;S

D; eappl

# $; (1)

where x is the 2D vector of position, g, t, D, S are the geo-metrical parameters introduced in Figure 1(b), and eappl isthe applied strain. We calculated the strain field by using thecommercial finite element software ABAQUS. An example ofthe calculated strain field in the three dimensional structureis shown in Figure 4(a). Both SU8 platforms and PDMS sub-strate were modeled as Neo Hookean material, with initial

FIG. 3. Surface strain mapping. (a)Experimental strain mapping at the topsurface of the engineered substrate. A20% macroscopic strain is applied alongthe x-axis. 50 lm thick, 1mm diameterSU8 platforms are embedded at the bot-tom of the PDMS membrane. The plat-forms’ spacing S varies from 2mm to10mm; (left) top view, optical imagesand (right) colored strain maps. (b)Strain profiles taken along the x-axisfrom the center of the SU8 platform. (c)Corresponding 3D, finite element, simu-lated profiles.

131904-3 Romeo et al. Appl. Phys. Lett. 102, 131904 (2013)

shear modulus and bulk modulus as 4GPa, 400GPa for SU8and 2MPa, 200MPa for PDMS. Symmetric boundary condi-tions (uniform displacement) are applied to any edge (sur-face for 3D case) running through the center of the platformin all three kinds of simulations showed in Figure 4. Such3D analysis is time-consuming, and is ineffective for study-ing the effect of geometric parameters. To simplify the anal-ysis, we next took advantage of a feature in our geometricdesign: the lateral dimensions are significantly larger thanthe thickness dimensions, i.e., D; S $ t; g. Consequently, the3D analysis (Figure 4(a)) can be well represented by usingtwo types of two-dimensional finite element analyses: top-view modeling (Figure 4(b)) and cross-section modeling(Figure 4(c)).

The top-view modeling is performed on a scale muchlarger than the thickness of the substrate (the global scale).We neglect the inhomogeneity across the thickness, and thestrain distribution takes the form,

ex

D

! "¼ fglobal

x

D;S

D; eappl

# $: (2)

This global-scale analysis was carried out using ABAQUS byassuming that the PDMS deforms under the plane-stress

conditions and that the SU8 platforms prescribe rigid bound-ary conditions (Figure 4(b)).

The cross-section modeling is performed in the PDMSat a distance from the edge of the rigid platform within a fewthickness of the substrate (the local scale). Here, the straindistribution is dominated by the heterogeneity and the localparameters, t; g, and the strain distribution takes the form,

ex

t

! "¼ flocal

x

t;g

t; efar

! ": (3)

This local strain, e xt

% &, is only function of the scalar, x, and

no longer on the vector x because the radius of curvature ofthe platform is significantly larger than t or g, and cannotinfluence the local strain distribution.

efar is the far-field strain, i.e., the strain prescribed to theright end of Figure 4(c). This strain is matched to the straincalculated at the edge of the platform from the global analy-sis. The local–scale analysis was carried out using ABAQUS byassuming that the PDMS deforms under the plane-strain con-ditions. The analysis resolves the detailed strain distributionwithin a cross section of PDMS (Figure 4(c)).

The elastomer is modeled as a neo-Hookean material.The material nonlinearity, however, is insignificant: our

FIG. 4. Computed strains. (a) 3D FEM simulation of a quarter of a 50lm thick, 1mm diameter SU8 platforms are embedded at the bottom of the PDMS mem-brane. The platform spacing is 5mm. Inset: localized, non-homogeneous strain distribution along the thickness. The 3D simulation can be decoupled in twocomplementary 2D models: (b) top view 2D in-plane stress model and (c) cross-section 2D in-plane stress model. (d) Strain profiles: the decoupled simulationscan catch the asymptotic behavior of the 3D simulation (S/D¼ 6). (e) Surface strain profile. (f) Cross-section model profiles. (g) Cross-sectional schematicillustrating the penetration depth. (h) Penetration depth as a function of the PDMS thickness above the platform.

131904-4 Romeo et al. Appl. Phys. Lett. 102, 131904 (2013)

simulation shows that the error due to material nonlinearityis usually less than 10%. The strain is approximately linearin the applied strain and the far field strain, so that theabove two equations can be written in simpler forms,

ex

D

! "¼ eappl % fglobal

x

D;S

D

# $(4)

and

ex

t

! "¼ efar % flocal

x

t;g

t

! ": (5)

The two functions were obtained from both finite-elementanalyses as described above.

We limit our presentation to the line where the higheststrain appears. This corresponds to the solid red line markedin Figure 4(b), where the load is applied from left to right.Figure 4(e) shows the calculation of fglobal x

D ;SD

% &, for the S/D

values corresponding to our experiments and compares theresults of the three kinds of finite-element models (Fig. 4(d)).As we expected, the local-scale analysis (Fig. 4(f)) can ap-proximate the real solution near the edge of the platformwhere the global-scale analysis (Fig. 4(e)) gives satisfactoryapproximation to the strain far away from the edge. Since theplatform is hardly deformable, the applied strain can only bedistributed over the area between the platforms; the larger theS/D ratio, the lower the average strain. There is stress concen-tration due to the presence of the stiff platform, where thesize of the influenced region is proportional to the size of theplatform. Figure 4(f) shows the calculation of flocal x

t ;gt

% &.

There is a strain peak due to the presence of the step of theplatform; the thicker the PDMS above the platform, thesmoother the transition. In the meantime, as the PDMS abovethe platform becomes thicker, the region on the platform thatis influenced by the far field strain gets larger (As decreases).

Although elastic wiring can sustain very large strains,we restrict our calculations to interconnect stretchability of20% strain and impose the As zone cannot sustain strainlarger than 0.5%. We define the penetration depth p as thedistance from the edge of the platform for which the straindecays to 0.5% if the far field strain is 20%, which gives thecriterion 0:025efar (Fig. 4(g)). So the factor 0.025 roughlyguarantees the maximal area usage of the platform. The pen-etration depth value is plotted in Figure 4(h); the result isalmost linear up till g/t¼ 0.6, above which the bending ofthe platform becomes significant. The linear relation showsthat approximately p ¼ 5g.

In summary, we have developed a simple and robustapproach to design and manufacture versatile elastic sub-strates for stretchable electronic applications. The geometricalconfiguration of the locally reinforced elastomer can be accu-rately optimized using two types of two-dimensional finite-element analyses prior to the microfabrication of the substrateusing standard UV photolithography and materials. We pro-pose simple guidelines on the geometry and density of the

rigid platforms embedded into the elastic matrix as well as onthe effective “safe” device island dimensions: (i) the deviceisland stack should be thin compared to the substrate, (ii) thestiff platform thickness should be selected so that the penetra-tion depth is minimized (p¼ 5 g at 20% applied strain) andthe strain at its edge is smooth, and (iii) the strain in-betweenthe stiff platforms decreases with increasing platforms density(S/D> 3). We have demonstrated that the engineered elasto-meric substrate is a promising carrier for thin-film devicematerials and metallization. Pre-stretching of the elastic sub-strate, i.e., buckled structures, is no longer required to mini-mize the strain in the most brittle advice device materials.Furthermore, the engineered stretchable substrate is preparedon a flat carrier that can then be handled similarly to standardmicroelectronics wafers yet provide the stretchability requiredonce released from the carrier. We demonstrate the potentialof patternable mechanical reinforcement of elastomer sub-strate with thin-film technology but believe this approach mayalso be compatible with transfer-printing and lamination ofelectronic circuits and components.

A.R. and S.P.L. acknowledge the support of an ERCStarting Grant No. 259419-ESKIN and the BertarelliFoundation. Q.L. and Z.S. acknowledge the support of theNational Science Foundation Materials Research Scienceand Engineering Center.

1T. Someya, Stretchable Electronics (Wiley-VCH, Weinheim, Germany,2013), p. 484.

2S. Wagner, S. P. Lacour, J. Jones, P.-H. Hsu, J. C. Sturm, T. Li, andZ. Suo, “13th International winterschool on new developments in solidstate physics, low-dimensional systems,” Physica E 25, 326–334 (2004).

3T. Someya, T. Sekitani, S. Iba, Y. Kato, H. Kawaguchi, and T. Sakurai,Proc. Natl. Acad. Sci. U.S.A. 101(27), 9966–9970 (2004).

4H. C. Ko, M. P. Stoykovich, J. Song, V. Malyarchuk, W. M. Choi, C.-J.Yu, J. B. Geddes III, J. X. Wang, Y. Huang, and J. A. Rogers, Nature 454,748–753 (2008).

5I. Graz, D. Cotton, A. Robinson, and S. P. Lacour, Appl. Phys. Lett. 98,124101 (2011).

6T. Someya, Y. Kato, T. Sekitani, S. Iba, Y. Noguchi, Y. Murase,H. Kawaguchi, and T. Sakurai, Proc. Natl. Acad. Sci. U.S.A. 102(35),12321–12325 (2005).

7T. Sekitani, H. Nakajima, H. Maeda, T. Fukushima, T. Aida, K. Hata, andT. Someya, Nature Mater. 8, 494–499 (2009).

8M. Gonzalez, F. Axisa, F. Bossuyt, Y.-Y. Hsu, B. Vandevelde, andJ. Vanfleteren, Circuit World 35(1), 22–29 (2009).

9J. A. Rogers, T. Someya, and Y. Huang, Science 327, 1603–1607 (2010).10D.-H. Kim, Z. Liu, Y.-S. Kim, J. Wu, J. Song, H.-S. Kim, Y. Huang, K.-C.Hwang, Y. Zhang, and J. A. Rogers, Small 5(24), 2841–2847 (2009).

11R.-H. Kim, D.-H. Kim, J. Xiao, B. H. Kim, S.-I. Park, B. Panilaitis, R.Ghaffari, J. Yao, M. Li, Z. Liu, V. Malyarchuk, D. G. Kim, A.-P. Le, R. G.Nuzzo, D. L. Kaplan, F. G. Omenetto, Y. Huang, Z. Kang, and J. A.Rogers, Nature Mater. 9, 929–937 (2010).

12S. P. Lacour, J. E. Jones, S. Wagner, T. Li, and Z. Suo, Proc. IEEE 93(8),1459–1467 (2005).

13I. Graz, D. Cotton, and S. P. Lacour, Appl. Phys. Lett. 94, 071902 (2009).14S. P. Lacour, D. Chan, S. Wagner, T. Li, and Z. Suo, Appl. Phys. Lett. 88,204103 (2006).

15Z. Suo, J. Vlassak, and S. Wagner, China Particuol. 3, 321 (2005).16G. Rochat, Y. Leterrier, P. Fayet, and J.-A. E. Manson, Thin Solid Films484, 94 (2005).

131904-5 Romeo et al. Appl. Phys. Lett. 102, 131904 (2013)